Family door lock data management method and system combined with large model

Abstract

The invention relates to the technical field of embedded development, in particular to a family door lock data management method and system combined with a large model, and the method comprises the following steps: initializing after a program is powered on, including initializing a display screen and keyboard input, and continuously reading the keyboard input until effective key operation is detected; the home door lock data management method and system combined with the large model, provided by the invention, have the beneficial effects that the home door lock data management method and system combined with the large model, pre-input is performed through fine adjustment of the large model and the outgoing time period, the common outgoing purpose and the spent time corresponding to the common outgoing purpose of the home members. According to the large model, firstly, the outgoing time period and the outgoing duration are combined, and secondly, different door opening passwords or door opening magnetic cards of the family members are combined, so that the outgoing family members and outgoing purposes are intelligently judged. And the understanding and concern of the old and children are enhanced.

Description

Technical Field

[0001] The present invention relates to the field of embedded development technology, in particular to a family door lock data management method and system combined with a large model. Background Art

[0002] Caring for the elderly and children and the safety of the elderly and children going out alone are the top priorities of family issues. In order to strengthen the care for the safety of the elderly and the supervision of children's behavior, the current methods are roughly as follows:

[0003] Children's watches: Children's watches are powerful and can make and receive calls, send and receive text messages, and locate children. However, there may be problems such as children forgetting to bring them or the battery running low during carrying. Overly detailed monitoring may cause children to be disgusted. Children can turn off these functions through technical means.

[0004] Locator: Traditional locators have the advantage of detailed monitoring and can accurately locate the position of children and the elderly. However, there may be problems such as the elderly and children forget to bring them or the battery is low during carrying. Similar monitoring methods may also cause children to be disgusted and resist. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for managing home door lock data in combination with a large model to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above object, the present invention provides the following technical solution: a method for managing home door lock data combined with a large model, the method comprising the following steps:

[0007] Initialization of the program after power-on, including initialization of the display screen and keyboard input, and continuous reading of keyboard input until valid key operations are detected;

[0008] Read the offline password stored in the EEPROM and compare it with the input password; if the password is wrong, the ERRORCIN count is increased, and after three consecutive errors, the keyboard is locked and an alarm is sounded; if the password is correct, the RIGHTCIN count is increased and the door lock is unlocked; when RIGHTCIN indicates that there has been a successful unlock, enter the password change mode again if the password is correctly entered;

[0009] In password modification mode, the user is required to enter the new password twice. If the two inputs are consistent, the new password will be written into the EEPROM; if they are inconsistent, the system will re-enter the initial mode;

[0010] Combined with the door lock status and time information, the door opening record is sent to the large model through the WiFi module for further processing;

[0011] Using the large model trained by LLaMA-Factory, we receive door lock data, monitor and analyze the behavior of family members going out / returning home, provide inference of the purpose of going out and chat suggestions, and send alarm information in abnormal situations.

[0012] Preferably, a large model training process is also included, including:

[0013] Data preparation step: collect travel data of family members of volunteers from multiple industries, write prompts, dialogues and questions and answers to form a training data set;

[0014] Model fine-tuning step: In LLaMA-Factory, the learning rate, batch size, and number of training rounds are adjusted according to the training dataset to optimize the model performance, and some parameters are modified to reshape the model's self-awareness function;

[0015] Evaluation and iteration steps: Test the fine-tuned model, evaluate its performance, and iterate and optimize it based on the feedback until it meets the actual application requirements.

[0016] Preferably, the large model pre-input and fine-tuning steps include:

[0017] Personalized information of family members, such as passwords, regular travel habits, etc., is input into the big model as a reference for model reasoning.

[0018] Preferably, the data monitoring and analysis step further includes:

[0019] Monitor family members' outings and homecomings in real time, combine historical data with current input to infer the purpose of going out, and provide personalized chat suggestions accordingly;

[0020] Set the alarm threshold. When it is detected that a family member is away for more than the preset time, an alarm message will be automatically sent to the homeowner.

[0021] Preferably, it also includes large model self-optimization learning, and the specific operations include:

[0022] Through the continuously accumulated home door lock usage data and homeowner feedback, the LLaMA-Factory tool is used to continuously train and optimize the large model to improve its recognition accuracy and the quality of output suggestions, thereby realizing intelligent and personalized home door lock data management.

[0023] A family door lock data management system combined with a large model, the system comprising:

[0024] Initialization module: used for initialization after the program is powered on, including initialization of the display screen and keyboard input, and continuous reading of keyboard input until a valid key operation is detected;

[0025] Password verification module: reads the offline password stored in the EEPROM and compares it with the input password. If the password is wrong, the ERRORCIN count is increased. After three consecutive errors, the keyboard is locked and an alarm is sounded. If the password is correct, the RIGHTCIN count is increased and the door lock is unlocked. When RIGHTCIN indicates that there has been a successful unlock, the password is entered again to enter the password modification mode.

[0026] Password modification module: In password modification mode, the user is required to enter a new password twice. If the two inputs are consistent, the new password will be written into the EEPROM; if they are inconsistent, the system will re-enter the initial mode;

[0027] Data collection and transmission module: Combined with the door lock status and time information, the door opening record is sent to the large model through the WiFi module for further processing;

[0028] Large model processing and analysis module: Utilizes the large model trained by LLaMA-Factory to receive door lock data, monitor and analyze the behavior of family members going out / returning home, provide inference of the purpose of going out and chat suggestions, and send alarm information in abnormal situations.

[0029] Preferably, a large model training module is also included, including:

[0030] Data preparation: Collect travel data of family members of volunteers from multiple industries, and write prompts, dialogues, and questions and answers to form a training data set;

[0031] Model fine-tuning: In LLaMA-Factory, adjust the learning rate, batch size, and number of training rounds according to the training dataset to optimize model performance, and modify some parameters to reshape the model's self-awareness function;

[0032] Evaluation and iteration: Test the fine-tuned model, evaluate its performance, and iterate based on the feedback until it meets the actual application requirements.

[0033] Preferably, the large model pre-input and fine-tuning includes:

[0034] Personalized information of family members, such as passwords, regular travel habits, etc., is input into the big model as a reference for model reasoning.

[0035] Preferably, data monitoring and analysis also includes:

[0036] Real-time monitoring of family members' going out and coming home behaviors, combining historical data and current input to infer the purpose of going out, and provide personalized chat suggestions accordingly; set alarm thresholds, and automatically send an alarm message to the head of the household when it is detected that a family member has been away for more than the preset time.

[0037] Preferably, a large model self-optimization learning module is also included, including:

[0038] Through the continuously accumulated home door lock usage data and homeowner feedback, the LLaMA-Factory tool is used to continuously train and optimize the large model to improve its recognition accuracy and the quality of output suggestions, thereby realizing intelligent and personalized home door lock data management.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] The family door lock data management method and system proposed by the present invention, combined with the big model, pre-inputs the time period of family members going out, common purpose of going out, and the time spent for common purpose of going out. The big model first combines the time period of going out and the duration of going out, and secondly combines the different door opening passwords or door opening magnetic cards of family members to intelligently judge the family members who are going out and the purpose of going out. Strengthen the understanding and care for the elderly and children.

[0041] If a family member is away for a long time, the large model will automatically send a text message to the homeowner to inform him or her that a family member has been away for a long time. The homeowner can choose to contact by phone or other options and means to confirm the safety of the elderly and children and ensure that safety hazards are discovered as early as possible.

[0042] By analyzing past data, for example, if the time an elderly person took to buy groceries today is much longer than the previous average, the system will automatically provide feedback: "The elderly person took a long time to buy groceries today. Maybe he met someone he knew. Go ask him~" This can increase common language and enhance family relationships. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a flow chart of the method of the present invention;

[0044] Figure 2 It is a flow chart of the data system of the present invention;

[0045] Figure 3 This is a flow chart of the door lock system of the present invention;

[0046] Figure 4 It is a schematic diagram of the door lock structure of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose and technical solution of the present invention clearly and completely described, and the advantages more clearly understood, the embodiments of the present invention are further described in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are part of the embodiments of the present invention, rather than all of the embodiments, and are only used to explain the embodiments of the present invention, and are not used to limit the embodiments of the present invention. All other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] Embodiment 1

[0049] See also Figures 1 to 4 The present invention provides a technical solution: a method for managing home door lock data combined with a large model, the method comprising the following steps:

[0050] Software and hardware selection:

[0051] 1. In terms of hardware, we first choose STC89C52 microcontroller as the carrier of the system. As the control center of the system, the specific password recognition, security mode setting and unlocking signal output are all executed by the microcontroller.

[0052] 2. The password lock part uses a 4x4 matrix keyboard, which occupies fewer interfaces and makes the system more compact.

[0053] 3. Added offline password modification function. AT24C02 is an EEPROM (Electrically Erasable Read-Only Memory), which is one of the key functions of the password lock and the key to modifying the password. Because EEPROM has the characteristic of not losing data after power failure, we use it to modify the password without modifying the code program, realizing more convenient human-computer interaction. And AT24C02 is very small.

[0054] 4. In order to better complete human-computer interaction, a display screen is used to guide visitors to the door. The main guidance is to guide the operation of the password lock, guide the input of the password, display the number of digits entered in the password, the password submission result, etc.

[0055] 5. Use batteries together with the power supply to ensure that the system will not be disconnected during a power outage.

[0056] 6. The large model uses the open source version of Qwen-8B, which has a small model and good performance, and can meet system requirements after fine-tuning.

[0057] 7. Add a wifi module. Every time the door lock is opened, the time information will be recorded and transmitted to the big model. After analysis and judgment by the big model, the output results will be written into the database.

[0058] Door lock setting process:

[0059] After the program is powered on, it will be initialized first, first the display screen and keyboard input will be initialized, and then the keyboard input will be read. If there is no input on the keyboard, it will keep reading. When the key is pressed, the interface level will change. After row and column scanning, the value of the key is obtained, and then different operations are performed according to the value of the key. Then it is the first submission. The system will read the offline access password in the EEPROM online. If the password is wrong, the ERRORCIN variable will be increased by 1. The function of this variable is that when the wrong password is entered three times in a row, the keyboard will be locked and the alarm will continue to sound; if the password is correct, the RIGHTCIN variable will be increased by 1, and then the door lock will be opened. The function of this RIGHTCIN variable is to record the number of correct entries. If the correct password is entered again after it has been successful once, the system will enter the password change mode.

[0060] After entering the modification mode, you need to reset the password and submit the password twice. After submission, if the two inputs are consistent, the password will be written into the offline register. If the two inputs are different, the initial mode will be re-entered.

[0061] Using LLaMA-Factory to train large models:

[0062] Step 1: Data Preparation

[0063] Collect travel information of family members of 100 volunteers from more than 30 industries, including catering, manufacturing, medium-sized enterprises, small and micro enterprises, and sales, to form basic data. Write prompts, dialogues, and questions and answers based on the basic data to form a data set.

[0064] Step 2: Fine-tune the model

[0065] In LLaMA-Factory, set appropriate training parameters, such as learning rate, batch size, number of training rounds, etc. By adjusting these parameters, the performance of the model can be optimized. According to the task requirements, some parameters of the model can be modified to reshape its self-awareness function.

[0066] Step 3: Evaluate and Iterate

[0067] After fine-tuning, we conduct tests to evaluate the accuracy and conduct further training and iteration to address defects.

[0068] Data system process:

[0069] Step 1: Pre-input and fine-tuning of large models

[0070] Feed family members with different passwords, frequently going out time periods, frequently going out purposes, duration of different going out purposes and other information for reference of the big model.

[0071] Step 2: Door lock settings

[0072] Set up passwords for different family members.

[0073] Step 3: Passing data

[0074] Based on the different sides of the door opening, it is determined whether the person is going out or coming home, and the door opening time is recorded and sent to the big model through the wifi module.

[0075] Step 4: Data monitoring and analysis

[0076] (1) Record out-of-town information and send it to the big model for monitoring. Combined with the out-of-town situation and out-of-town time period, analyze which family member is most likely to be out-of-town.

[0077] (2) Record the return information and send it to the big model for judgment. Combined with the members' outings, the time period when they return home, the duration of their outings, and the door opening password, the model comprehensively analyzes which family member is most likely to return home, infers the purpose of their outing, and gives chat suggestions.

[0078] (3) If a family member is away for too long, an alarm message will be sent to the homeowner.

[0079] Step 5: Self-optimization learning of large models

[0080] Through the accumulation of information and fine-tuning of data by the owner, the LLaMA-Factory tool is used to increase the accuracy of large model recognition and output better chat suggestions.

[0081] Embodiment 2

[0082] Based on the first embodiment, a family door lock data management system combined with a large model is proposed, the system comprising:

[0083] Initialization module: used for initialization after the program is powered on, including initialization of the display screen and keyboard input, and continuous reading of keyboard input until a valid key operation is detected;

[0084] Password verification module: reads the offline password stored in the EEPROM and compares it with the input password. If the password is wrong, the ERRORCIN count is increased. After three consecutive errors, the keyboard is locked and an alarm is sounded. If the password is correct, the RIGHTCIN count is increased and the door lock is unlocked. When RIGHTCIN indicates that there has been a successful unlock, the password is entered again to enter the password modification mode.

[0085] Password modification module: In password modification mode, the user is required to enter a new password twice. If the two inputs are consistent, the new password will be written into the EEPROM; if they are inconsistent, the system will re-enter the initial mode;

[0086] Data collection and transmission module: Combined with the door lock status and time information, the door opening record is sent to the large model through the WiFi module for further processing;

[0087] Large model processing and analysis module: Utilizes the large model trained by LLaMA-Factory to receive door lock data, monitor and analyze the behavior of family members going out / returning home, provide inference of the purpose of going out and chat suggestions, and send alarm information in abnormal situations.

[0088] It also includes large model training modules, including:

[0089] Data preparation: Collect travel data of family members of volunteers from multiple industries, and write prompts, dialogues, and questions and answers to form a training data set;

[0090] Model fine-tuning: In LLaMA-Factory, adjust the learning rate, batch size, and number of training rounds according to the training dataset to optimize model performance, and modify some parameters to reshape the model's self-awareness function;

[0091] Evaluation and iteration: Test the fine-tuned model, evaluate its performance, and iterate based on the feedback until it meets the actual application requirements.

[0092] Model pre-input and fine-tuning include:

[0093] Personalized information of family members, such as passwords, regular travel habits, etc., is input into the big model as a reference for model reasoning.

[0094] Data monitoring and analysis also includes:

[0095] Real-time monitoring of family members' going out and coming home behaviors, combining historical data and current input to infer the purpose of going out, and provide personalized chat suggestions accordingly; set alarm thresholds, and automatically send an alarm message to the head of the household when it is detected that a family member has been away for more than the preset time.

[0096] Also includes large model self-optimization learning modules, including:

[0097] Through the continuously accumulated home door lock usage data and homeowner feedback, the LLaMA-Factory tool is used to continuously train and optimize the large model to improve its recognition accuracy and the quality of output suggestions, thereby realizing intelligent and personalized home door lock data management.

[0098] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for managing home door lock data in combination with a large model, characterized in that: The method comprises the following steps: Initialization of the program after power-on, including initialization of the display screen and keyboard input, and continuous reading of keyboard input until valid key operations are detected; Read the offline password stored in the EEPROM and compare it with the input password; if the password is wrong, the ERRORCIN count is increased, and after three consecutive errors, the keyboard is locked and an alarm is sounded; if the password is correct, the RIGHTCIN count is increased and the door lock is unlocked; when RIGHTCIN indicates that there has been a successful unlock, enter the password change mode again if the password is correctly entered; In password modification mode, the user is required to enter the new password twice. If the two inputs are consistent, the new password will be written into the EEPROM; if they are inconsistent, the system will re-enter the initial mode; Combined with the door lock status and time information, the door opening record is sent to the large model through the WiFi module for further processing; Using the large model trained by LLaMA-Factory, we receive door lock data, monitor and analyze the behavior of family members going out / returning home, provide inference of the purpose of going out and chat suggestions, and send alarm information in abnormal situations.

2. A method for managing home door lock data in combination with a large model according to claim 1, characterized in that: It also includes the large model training process, including: Data preparation step: collect travel data of family members of volunteers from multiple industries, write prompts, dialogues and questions and answers to form a training data set; Model fine-tuning step: In LLaMA-Factory, the learning rate, batch size, and number of training rounds are adjusted according to the training dataset to optimize the model performance, and some parameters are modified to reshape the model's self-awareness function; Evaluation and iteration steps: Test the fine-tuned model, evaluate its performance, and iterate and optimize it based on the feedback until it meets the actual application requirements.

3. A method for managing home door lock data in combination with a large model according to claim 2, characterized in that: The steps of pre-input and fine-tuning of large models include: Personalized information of family members, such as passwords, regular travel habits, etc., is input into the big model as a reference for model reasoning.

4. A method for managing home door lock data in combination with a large model according to claim 3, characterized in that: The data monitoring and analysis steps also include: Monitor family members' outings and homecomings in real time, combine historical data with current input to infer the purpose of going out, and provide personalized chat suggestions accordingly; Set the alarm threshold. When it is detected that a family member is away for more than the preset time, an alarm message will be automatically sent to the homeowner.

5. A method for managing home door lock data in combination with a large model according to claim 4, characterized in that: It also includes large-scale model self-optimization learning, and the specific operations include: Through the continuously accumulated home door lock usage data and homeowner feedback, the LLaMA-Factory tool is used to continuously train and optimize the large model to improve its recognition accuracy and the quality of output suggestions, thereby realizing intelligent and personalized home door lock data management.

6. A home door lock data management system combined with a large model according to the home door lock data management method combined with a large model according to any one of claims 1 to 5, characterized in that: The system comprises: Initialization module: used for initialization after the program is powered on, including initialization of the display screen and keyboard input, and continuous reading of keyboard input until a valid key operation is detected; Password verification module: reads the offline password stored in the EEPROM and compares it with the input password. If the password is wrong, the ERRORCIN count is increased. After three consecutive errors, the keyboard is locked and an alarm is sounded. If the password is correct, the RIGHTCIN count is increased and the door lock is unlocked. When RIGHTCIN indicates that there has been a successful unlock, the password is entered again to enter the password modification mode. Password modification module: In password modification mode, the user is required to enter a new password twice. If the two inputs are consistent, the new password will be written into the EEPROM; if they are inconsistent, the system will re-enter the initial mode; Data collection and transmission module: Combined with the door lock status and time information, the door opening record is sent to the large model through the WiFi module for further processing; Large model processing and analysis module: Utilizes the large model trained by LLaMA-Factory to receive door lock data, monitor and analyze the behavior of family members going out / returning home, provide inference of the purpose of going out and chat suggestions, and send alarm information in abnormal situations.

7. A household door lock data management system combined with a large model according to claim 6, characterized in that: It also includes large model training modules, including: Data preparation: Collect travel data of family members of volunteers from multiple industries, and write prompts, dialogues, and questions and answers to form a training data set; Model fine-tuning: In LLaMA-Factory, adjust the learning rate, batch size, and number of training rounds according to the training dataset to optimize model performance, and modify some parameters to reshape the model's self-awareness function; Evaluation and iteration: Test the fine-tuned model, evaluate its performance, and iterate based on the feedback until it meets the actual application requirements.

8. A household door lock data management system combined with a large model according to claim 7, characterized in that: Large model pre-input and fine-tuning include: Personalized information of family members, such as passwords, regular travel habits, etc., is input into the big model as a reference for model reasoning.

9. A household door lock data management system combined with a large model according to claim 8, characterized in that: Data monitoring and analysis also includes: Real-time monitoring of family members' going out and coming home behaviors, combining historical data and current input to infer the purpose of going out, and provide personalized chat suggestions accordingly; set alarm thresholds, and automatically send an alarm message to the head of the household when it is detected that a family member has been away for more than the preset time.

10. A household door lock data management system combined with a large model according to claim 9, characterized in that: Also includes large model self-optimization learning modules, including: Through the continuously accumulated home door lock usage data and homeowner feedback, the LLaMA-Factory tool is used to continuously train and optimize the large model to improve its recognition accuracy and the quality of output suggestions, thereby realizing intelligent and personalized home door lock data management.

Images